Atomic data-storage device developed

The device is the largest atomic structure ever built by humans and has an impressive information density of up to 500 terabits per square inch.

Sander Otte, Delft University of Technology and colleagues developed a binary alphabet based on atomic vacancies on a copper surface coated with chlorine atoms. In the paper published in Nature Nanotechnology they demonstrate reliable write, read-out and rewrite operations in a 1-kilobyte device consisting of 8,000 chlorine vacancies (missing atoms). The research team managed to preserve a similar but smaller atomic device for more than 40 hours at 77 Kelvin. They stored information including Richard P. Feynman’s lecture ‘There’s plenty of room at the bottom’ on the device which also inspired the research.

ResearchGate: Can you explain your study? What have you found?

Sander Otte: We have been able to store a kilobyte of data (8,000 bits) in such a way that each bit is represented by the position of a single atom. For this purpose, we use atom manipulation by means of a scanning tunneling microscope (STM). But in contrast to many previous studies that involved positioning of a few widely dispersed individual atoms on a surface, in our case we have a surface that is almost saturated with chlorine atoms and we move around the missing atoms (vacancies) instead. You can best compare this to one of those sliding puzzles in which one piece is missing.

The key advantage of this approach is that the almost saturated chlorine layer provides a beautiful smooth square grid that serves as the template for our memory. On each site in the grid there can either be an atom or not, so it’s perfectly digital. This not only allows us to clearly define the zeroes and ones of our memory, it also enables us to write atomic markers that can be read out by the STM tip so that it can navigate autonomously through the memory. Examples of information encoded in the markers are “you have reached the end of the line; go back 100 nm and then 10 nm south to find the beginning of the next line”, or “this area is contaminated and cannot be used for data storage; please proceed to the next sector”. The implementation of these kinds of protocols are essential for scaling up the technology.

The final memory we built contained 8,000 atomic vacancies, or alternatively over 60,000 chlorine atoms, most of which were manipulated at least once.



RG: Is the idea of an atomic storage device new? What inspired this line of research?

Otte: Not at all. I believe the credit for starting this line of research, at least on a conceptual level, should go to Richard Feynman. For this reason, we encoded a part of his lecture “There’s plenty of room at the bottom” as the first text in our memory.

There have been several experimental efforts in this direction as well, of which I would like to mention two. First, in 2002 a group from the University of Madison Wisconsin (Bennewitz et al., Nanotechnology 2002) demonstrated storage of information in the presence (or not) of individual Si atoms. While this was an excellent achievement for that time, the memory was unfortunately not rewritable: once an atom was shot away, it could not be put back.

Second, in 2012 Loth et al. reported in Science the construction of a stable magnetic byte built out of individual iron atoms. Each bit in this byte consisted of 12 atoms. From a physics perspective that was a beautiful experiment, because it showed the transition from a single quantum spin (which is in a superposition) to a collection of spins that together behave as a classical magnet. I would call it the ‘birth’ of classical magnetism, and this had never been witnessed before.

Compared to that last work, our current data storage in the positions of atoms is of course rather mundane. But in terms of reliability, stability and scalability it is a tremendous step forward.

RG: What was the most challenging aspect of this study?

Otte: This may sound presumptuous, but once we had figured out the proper settings for moving the vacancies, building the large memory was surprisingly easy. I suppose the most challenging (and fun) part was the pathfinding of the vacancies: what is the most efficient way to turn a random distribution of vacancies into a bit array? Most of the time our algorithms worked well, but sometimes we saw the STM do really weird things: it would for example drag a single vacancy all around a nearly finished block rather than moving over each vacancy in a row to make space for the last one. But in the end the manipulation worked so well that we just let the STM do what it wanted, as long as the final result was good.

RG:  If these results are “proof-of-principle” what are the next steps?

Otte: In terms of optimizing data storage, a next step would be to try to improve stability to even higher temperatures. The chlorine vacancies we used are stable up to 77 K, but by trying out similar but different material combinations we might be able to increase this. Another obvious step is speeding up the read/write process, which is currently still extremely slow. Physically I foresee no limitation in making this as fast as HDD read/write speeds (say 1 Mb/s) for example, but technologically there are some serious challenges. For example, when a piezo-element (which moves the STM tip) is driven very fast, its response becomes non-linear. It will require a lot of calibration to get that to work well.

STM scan (96 nm wide, 126 nm tall) of the 1 kB memory, written to a section of Feynman’s lecture 'There’s Plenty of Room at the Bottom' (with text markup). Image courtesy of TU Delft
STM scan (96 nm wide, 126 nm tall) of the 1 kB memory, written to a section of Feynman’s lecture 'There’s Plenty of Room at the Bottom' (with text markup). Image courtesy of TU Delft

RG: Why are data storage innovations like this so important?

Otte: The amount of data we store as a society continues to grow rapidly, consuming lots of space and energy. Yet, as Feynman pointed out, there is plenty of space available if we know how to access it. It would be a waste not to explore all that space.

RG: How long will it take before this type of storage device could be available to the public? What type of uses do you envision for it?

Otte: First of all, I would say that the only way I envision this to become practical would be in a centralized data storage environment (a data center), and not as a local storage solution like the one on your smartphone. Yet, even for that application we are still far off and I would caution people not to get excited about applications too quickly.

But instead of focusing on applications, I would like to emphasize that in my view the true importance of this work is not necessarily the data storage itself. This large kilobyte memory should rather be seen as a demonstration of how well we can now organize the world with atomic precision. Who knows what possibilities will arise once we can start to design and build our own functional nanomaterials rather than being limited to the materials that nature provides us. Or what clever new quantum technologies we may come up with when we can create all kinds of circuits on the atomic scale. I cannot at this point foresee where this will lead to, but I am convinced that it will be much more exciting than just data storage.

Explanation of the bit logic and the atomic markers. Image courtesy of TU Delft
Explanation of the bit logic and the atomic markers. Image courtesy of TU Delft

RG: What are you thinking of in this regard?

Otte: I would like to encourage everyone to use their imagination and invent new experiments and technologies that have now become possible. As we describe in our paper, the recipe for doing this kind of large-scale manipulation is not at all demanding. There are many groups in the world that possess the know-how and the equipment to reproduce our work. We may currently hold the world-record for having built the largest atomic assembly ever, but I truly hope that this record will be broken very soon.

Featured image courtesy of Christiaan Colen.